Week 4 Notes

Notes

WHY use Spatial Analysis

• WHERE are these patterns occurring matters
  • Geographic clustering of problems
  • Spatial relationships between communities
  • Access to services and resources is often based on geography

Spatial Data Fundamentals

• Two families of representing world 2-dimensionally
  • Vector
    • discrete objects – things that have definite boundaries
  • Raster
    • Pixels
    • Continuous data

Vector Data Representation - Three basic types of geometric representations - points - lines - polygons - (kinda like illustrator)

Common Spatial Data File Formats - Shapefile - Developed by ESRI - Three fundamental objects (need all three to render accurately): - .shp - stores info about the geometry - .shx - shape index - .dbf - names of things - Integrates with tidyverse, follows inernational standards - GeoJSON - All one file ! Little more - KML - Google Earth - Database connections (PostGIS)

Simple Features - Multi-shapes – like a broken bridge – think multiple shapes for one row of a table - Think Hawaii - -multiple shapes that are all “Hawaii”

TidyCensus gives you characteristics but no shapes, tigris gives you shapes but no characteristics – must combine

Important syntax:

• ==ggplot geom for mapping: geom_sf()==
• ==ggplot use theme_void() to get rid of graph backdrop==
• ==st_filter() is like filter() but for spatial filtering rather than simple df`
• st_union() works like arcgis “dissolve”

Spatial Subsetting: - preserves original shapes, not for doing something like finding part of a county. Rather for doing things like finding a county within a list of counties, or all counties that touch a specific county

neighbors <- pa_counties %>% st_filter(allegheny, .predicate = st_touches)

• In this example, the terminology “st_touches” is common across all GIS
• .predicate tells st_filter() what kind of relationship to look for, if nothing specified, then st_intersect is default

cheat sheet:

cheat-sheet

Coordinate Reference System (CRS)

• basic problem : world is round, maps are flat, projecting from 3d to 2d will necessarily cause some kind of distortion
• Earth is also not perfectly round, but a geoid ( a lil bumpy ), but geoid is mathematically inconvenient, so the Step 1 in projection is we project from he geoid to an ==ellipsoid==, which doesn’t perfectly represent the geoid, but good enough, and smooth
  • This means there are multiple ellipsoids that fit the earth, and might fit certain areas better than others

crs

• Step 2 – tie the ellipsoid to the real earth to create a Geographic (Geodetic) Coordinate System i.e. Lat Longs
  • Clark, 1866, uses ‘flattening’ to make a nice lil ellipsoid that is particularly well-fitted to North America – Meades Ranch, Kansas, is where the ellipsoid and geoid smooch :-*
  • ==North American Datum 1927== or ==NAD27==
  • Progress made, made a better ellipsoid than Clarke (“We’re not in Kansas anymore”), based on Earth Center instead of NA.. this one called ==GRS80==
  • ==WGS84== – yet another used by GPS systems
• Step 3 – take the 3d ellipsoid points and project onto 2d surface
  • Cylindrical Projections

cylindical-projections

    -   Line of tangency -- where the ellipsoid and the projected
        surface smooch, all other locations will be distorted
    -   Mercator is prime example
    -   Bad at preserving sizes of things on 2d surface, but good at
        preserving angles
-   Transverse Cylindrical Projection
    -   Good at preserving up and down (good for Chile, for example)
    

transverse

-   Conical Projection
    -   Good for areas concentrated in a segment of the ellipsoid,
        like North America
        

conical

-   Projected Coordinate System
    -   Localized coordinate system built on a regular,
        non-distorted grid
    -   ==UTM== --

UTM

        -   Need to know what zone you are in -- all numbers are
            based on grid system originating in the lower left
            corner of the zone
        -   No negatives
        -   Who uses UTM ? depends on state -- if looking at a long
            skinny area like Idaho for example
    -   ==State Plane== (SPC)
        -   Used in PA - Two state plane grids in PA, North and
            South

state-plane

        -   Some states use Feet, others use Meters -- make sure to
            check !
-   SADD -- can't have all 4 in a projection
    -   Shape
    -   Area
    -   Distance
    -   Direction

• st_crs() to check CRS for a dataset
• st_set_crs(data, ####) – set CRS – ONLY IF CRS MISSING, do NOT use to transform CRS (path of despair) – this will not change the numbers themselves but rather how the computer interprets the numbers, only use if st_crs() returns <unknown>
• st_transform() – this will actually recalculate the coordinates from one CRS to another
• In arcGIS – this is the like the difference between “define projection” and “project”